DEM estimation from multi-Baseline ENVISAT- ASAR interferometric data through maximum likelihood techniques

In this paper, two techniques for estimating accurate height profiles of the ground, using multi-baselines interferometric synthetic aperture radar (In-SAR) data and an a-priori inaccurate digital elevation model (DEM) of the observed scene, are analyzed. The methods are both based on maximum likelihood (ML) estimation: the first estimates directly the quota of each pixel of the image, independently from the other pixels, while the latter estimates the parameters of the local planes which best approximate, in the ML sense, the height profile in a small neighborhood of each pixel. The inclusion of this contextual information allows improving the estimation accuracy. Results on simulated and real ENVISAT-ASAR data are presented.